We screened anonymously all mothers and infants born during a 3 1/2-month period to determine the prevalence of intrapartum cocaine use, test the maternal characteristics that are specific predictors of intrauterine cocaine exposure (IUCE), and compare the sensitivity of infant urine versus meconium samples for identification of IUCE. Of 1237 live births during the study period, a sample was obtained from 1201 mother-infant pairs. The overall prevalence of documented intrapartum cocaine exposure was 66 (5.5%) of 1201 pairs. Previously developed drug screening guidelines had a sensitivity of 89% for detecting IUCE in infants. Direct comparisons of samples from the same mother-infant pair revealed that there were no cases in which cocaine was found in infant urine but not in meconium; however, infant urine testing missed 25% of the infants who had positive findings in meconium. We conclude that (1) meconium testing was more likely than urine testing to identify an infant with IUCE, detecting an additional 33%; (2) there was significant maternal cocaine use (5.5%) in a teaching hospital with a mixed patient population; (3) maternal characteristics known to identify infants at risk of having IUCE were useful in our population; and (4) IUCE of neonates admitted to the neonatal intensive care unit was more common than that of infants admitted to the regular newborn nursery.